Fit a survival causal forest workflow

Usage

fit_survival_forest(
  data,
  time = "time",
  event = "event",
  treatment = "treatment",
  covariates,
  horizon,
  target = c("RMST", "survival.probability"),
  sample_id = NULL,
  candidate = NULL,
  num_trees = 2000,
  min_node_size = 5,
  tree_depth = 3,
  tree_minbucket = 100L,
  tree_trim_quantiles = c(0.05, 0.95),
  seed = NULL
)

Arguments

data: A single analysis data.frame.
time: Observed event or censoring time column.
event: Event indicator column. Use 1 for event and 0 for censoring.
treatment: Treatment assignment column.
covariates: Baseline covariates for confounding adjustment and heterogeneity discovery.
horizon: Horizon used by grf::causal_survival_forest().
target: Survival estimand. Either "RMST" or "survival.probability".
sample_id: Optional sample identifier column.
candidate: Optional treatment-comparison label column.
num_trees: Number of trees for grf::causal_survival_forest().
min_node_size: Minimum node size for grf::causal_survival_forest().
tree_depth: Maximum depth of the explanation tree.
tree_minbucket: Minimum leaf size of the explanation tree.
tree_trim_quantiles: Quantiles used to clip extreme effect estimates before fitting the explanation tree.
seed: Optional random seed.

Value

A heteff_fit object.